Re: segfault in scm_gc_mark?

[Greg Harvey <Greg dot Harvey at thezone dot net>]

Lynn Winebarger <owinebar@free-expression.org> writes:


>     I've updated to the CVS version (last night), and compiled with
> debugging turned on.  Now it recognizes the whole stack, which is helpful. 
>     What kinds of objects are you referring to?  I can't see anything in
> my code that isn't properly initialized, at least in terms of what I've
> malloc'ed (which is not much - most of the mallocing is done while loading
> the binary and symbol/relocation tables).  I don't use any smobs - just
> gh_cons, gh_symbol2scm (which isn't really a heap allocation anyway), and
> a few C->scheme value converters.  

It could be that you're putting incorrect values somewhere (for
example, seting a guile object to an integer, rather than
SCM_INUM(integer)). Or you could be accidentally overwriting values
somewhere (this is where checking what the value causing the segfault
comes in handy; it might look sort of familiar). The simple objects
here (cons cells, symbols [it does do some heap allocation, btw... a
few cells]) shouldn't have any gc segfaulting problems, though it is
possible.

-- 
Greg